Features

A serious AI chat, not just a private one.

Privacy is the foundation, not the feature list. On top of it you get a capable, modern chat: reasoning modes, model choice, tools, receipts, and dedicated deployments for teams.

Choose how it answers.

Choose your model.

Privasys Chat runs open-weight models: models whose weights are public, so what you are running can actually be verified. That verifiability is the point; a black-box API could never carry a checkable fingerprint.

Each instance offers a menu of models in the message bar, and the platform can serve any open-weight model, from compact fast responders to large reasoning models. If the model your team needs is not on the menu, we can host it for you. The current line-up and hardware details are in the models documentation.

Your own private instance.

The public instance is shared infrastructure with private sessions. Teams with stronger requirements get a dedicated deployment of their own.

AI tools, without leaving the boundary.

The model can use tools, such as web search or a browser, and those tools run inside attested enclaves too. Each tool gets its own row in the Security panel, verified like the model itself.

You choose which tools are active per conversation from the message bar. Depending on the instance policy, you can add your own tools; anything that would leave the protected boundary is clearly marked as unverified and requires your explicit acknowledgement.

The details that add up.